murray@minster.york.ac.uk (Kevin Murray) (02/09/88)
Wisdom is a Distributed Operating System that is just entering its implementation phase (with some design still to be done). It is intended to run on an array of processors, initially targetted at transputers. It is to be implemented in an augmented version of occam. The system design is one of a minimal set of modules (routing, load balancing and naming) providing the necessary services for the system, with all of them (hopefully) dynamically replaceable. The extensions currently under consideration for occam are to allow dynamic processes (hence load balancing) and to allow CHANnels to be passed over CHANnels (hence routing). All other services (e.g. filesystems, device drivers) are viewed as user processes and would be implemented as such. The resultant model of the system is one of dynamic objects (processes) communicating dynamically: i.e. object oriented model, with the hardware layout hidden, when necessary, by the ``operating system''. Finally it is possible that I may get a travel grant allowing me to spend some time abroad visiting several research institutions, so if anyone out there has interest in the work I am doing, or feels that they have work which is relevant, could the contact me please. ------------ Kevin Murray, Dept. of Comp. Sci., University of York, York, UK, YO1 5DD "A dead body can be very difficult to explain, especially if it's your own" -- Rod Gallowglass, A Wizard in Bedlam by Christopher Stasheff "I'm going to live forever, or die trying" -- Vila, ??, Blake's Seven murray@minster.york.ac.uk or ..!uunet!mcvax!ukc!minster!murray
murray@minster.york.ac.uk (02/24/88)
>Frankly it doesn't sound like Wisdom is much of an operating system. I I admit that in the current traditional mode it isn't. What should be borne in mind is that operating systems tend to be getting smaller (IBM excepting!). By that I mean less and less is being done in the ``kernel'' and more and more is being done by ``user'' processes. >would say it soulds much more like a monitor given the lack of support >and protection such a loose organization implies. All of the protection needed (well almost all) is afforded by the semantics of occam, which at the moment is the only language that you'll be able to use on it. It may well appear like a monitor in some respects, but I think that is wrong (suprisingly!). What I hope I've got is the minimal subset of ``services'' that will allow anything else to be built on top of them, with at least a reasonable degree of ease. One comparison that could be drawn is with Codd's Database requirements of a query language, which allow you to do almost anything. Possibly both with similar drawbacks. > >- John. >--- >John F. Haugh II SNAIL: HECI Exploration Co. Inc. >UUCP: ...!ihnp4!killer!jfh 11910 Greenville Ave, Suite 600 >"Don't Have an Oil Well? ... Dallas, TX. 75243 > ... Then Buy One!" (214) 231-0993 Ext 260 > Kevin ---- Kevin Murray, Dept. of Comp. Sci., University of York, York, UK, YO1 5DD "A dead body can be very difficult to explain, especially if it's your own" -- Rod Gallowglass, A Wizard in Bedlam by Christopher Stasheff "I'm going to live forever, or die trying" -- Vila, ??, Blake's Seven murray@uk.ac.york.minster or ..!uunet!mcvax!ukc!minster!murray
martin@felix.uucp (Martin McKendry) (03/03/88)
In article <4668@sdcsvax.UCSD.EDU> murray@minster.york.ac.uk writes: >>Frankly it doesn't sound like Wisdom is much of an operating system. I > >I admit that in the current traditional mode it isn't. What should be >borne in mind is that operating systems tend to be getting smaller (IBM >excepting!). By that I mean less and less is being done in the ``kernel'' >and more and more is being done by ``user'' processes. I have to take exception to this. Academic and research operating systems seem to think small kernels are a wonderful goal. This is fine. It makes the system easier to add to, and to test. However, there is a cost in efficiency. Industrial kernels are getting bigger and bigger for this reason. In my opinion, a fruitful area of research is how to organize large kernels. There is nothing inherently wrong with being big (where big is, say, order of 500 KB). We have the memory for it. The only thing a 'small' kernel does is provide tighter protection boundaries, at the cost of efficiency and less application-oriented kernel interfaces. And you still have to hire operating system engineers to build the "user" processes that do such functions as maintaining networks and file systems. -- Martin S. McKendry; FileNet Corp; {hplabs,trwrb}!felix!martin Strictly my opinion; all of it